LKB1 associates with Brg1 and is necessary for Brg1-induced growth arrest.

Inactivating mutations in the serine-threonine kinase LKB1 (STK11) are found in most patients with Peutz-Jeghers syndrome; however the function of LKB1 is unknown. We found that LKB1 binds to and regulates brahma-related gene 1 (Brg1), an essential component of chromatin remodeling complexes. The association requires the N terminus of LKB1 and the helicase domain of Brg1 and LKB1 stimulates the ATPase activity of Brg1. Brg1 expression in SW13 cells induces the formation of flat cells indicative of cell cycle arrest and senescence. Expression of a kinase-dead mutant of LKB1, SL26, in SW13 cells blocks the formation of Brg1-induced flat cells, indicating that LKB1 is required for Brg1-dependent growth arrest. The inability of mutants of LKB1 to mediate Brg1-dependent growth arrest may explain the manifestations of Peutz-Jeghers syndrome.

, Par4 Caenorhabditis elegans partitioning-defective gene 4 (12), and a Drosophila melanogaster gene (accession number AAF54972) are LKB1 orthologs, but little is known about the function of any of the proteins. Par4 is required for asymmetric cell division of the C. elegans embryo, but its specific role is not understood (12). XEEK1 is expressed early in Xenopus development, phosphorylates a protein of 155-kDa, and is a substrate for cAMP-dependent protein kinase (11). LKB1 can also be phosphorylated by cAMP-dependent protein kinase, as well as p90 Rsk (13,14). XEEK1 localizes exclusively to the cytoplasm (11), but LKB1 is found in both the nucleus and the cytoplasm (15). The SL26 mutant of LKB1, isolated from a patient with PJS (16), is localized almost exclusively to the nucleus (15,17). Expression of LKB1 in G361 melanoma cells resulted in significant inhibition of cell growth because of a G 1 arrest, suggesting that LKB1 may be involved in cell cycle regulation (14,18). To understand the function of LKB1, we carried out a screen to identify proteins to which it binds. We found that LKB1 associates with Brg1, a component of SWI⅐SNF chromatin remodeling complexes. LKB1 stimulates the ATPase activity of Brg1 and is required for Brg1induced growth arrest.
Plasmids-Brg1⌬HpcDNA3.1 was constructed by digesting Brg1HRpcDNA3.1 with SmaI, and the resulting fragment was ligated into pcDNA3.1(ϩ) at EcoRV and SmaI. Brg1HpcDNA3.1 was constructed by introducing a BamHI site and an XhoI site followed by ligation into pcDNA3.1(ϩ) at BamHI and XhoI. Brg1HpGEX4T2 and Brg1⌬HpGEX4T2 were constructed by digesting Brg1HpcDNA3.1 and Brg1⌬HpcDNA3.1 with BamHI and XhoI and ligating into pGEX4T2 at BamHI and XhoI. Brg1HRpGEX4T2 was constructed by digesting Brg1HRpcDNA3.1 with BamHI and XhoI followed by ligation into pGEX4T2 at BamHI and XhoI. LKB1pGEX4T2 and SL26pGEX4T2 were constructed by digesting LKB1 and SL26 with EcoRI and SalI from LKB1pAHC and SL26pAHC followed by ligation into pGEX4T1 at EcoRI and SalI. ⌬C-LKB1pGEX4T1 was constructed by digesting LKB1pAHC with EcoRI and ScaI followed by ligation into pGEX4T1 at EcoRI and SmaI. ⌬N-LKB1 was constructed by digesting LKB1 with BstXI followed by Klenow dNTP addition and SalI digestion and ligation into pGEX4T2 at SmaI-SalI. LKB1pBJ5 and SL26pBJ5 were constructed by digesting LKB1pAHC and SL26pAHC with XhoI and NotI followed by ligation into pBJ5 at XhoI and NotI.
Kinase Assay-GST-LKB1 and GST-SL26 were purified from insect cells using GSH-agarose beads. The proteins were eluted from the beads with GSH, concentrated, and then diluted to remove the GSH. Kinase assays were conducted at 30°C for 30 min in kinase buffer (20 mM Hepes, pH 7.5, 5 mM MgCl 2 , 0.4 mM MnCl 2 , 1 mM dithiothreitol, 75 g/ml bovine serum albumin, 20 M ATP, and 10 Ci of [␥-32 P]ATP). Samples were separated by SDS-PAGE, and radioactivity was visualized by autoradiography.
Flat Cell Assay-2 ϫ 10 6 SW13, from primary small cell lung carcinoma from an adrenal metastasis (ATCC number CCL-105), were transiently transfected with Brg1, LKB1, SL26, pBJ5 vector alone, or Brg1 ϩ LKB1 or SL26. pcDNA3.1 containing a neomycin resistance gene was included at a ratio of 5:1 for selection in neomycin. Beginning 48 h after transfection, cells were treated with neomycin (200 g/ml) for an additional 14 days. Cells were washed three times in phosphate-buffered saline, fixed in 3% paraformaldehyde, washed, and then stained with 2% crystal violet. For each condition, the number of flat cells in 40 randomly selected 40ϫ fields were counted.

RESULTS AND DISCUSSION
To identify proteins that interact with LKB1, we adapted an in vitro expression cloning strategy (19) in which a HeLa cell cDNA library was divided into 680 pools of ϳ100 cDNAs per Labeled proteins were incubated with GST-LKB1 (10 pmol) or GST (10 pmol) and washed, and bound proteins were then separated by SDS-PAGE and visualized using a molecular imager. The arrow indicates the 35-kDa protein that associates specifically with GST-LKB1, following isolation of the clone. 14 C-MW, 14 Clabeled molecular mass markers. B, LKB1 was immunoprecipitated from Saos-2 cells using LKB1 antibody (␣-LKB1), and the immunoprecipitates were Western blotted for LKB1 and Brg1. Pre-immune serum was used as a control (PIS), and total cell lysates (TCL) were also blotted for LKB1 and Brg1. A, Brg1 ATPase activity was assayed in the presence and absence of DNA and following the addition of GST-LKB1, GST-SL26, or GST proteins as described (20). The increase in activity because of LKB1 or SL26 was statistically significant (*, p Ͻ 0.0001, and **, p Ͻ 0.005, respectively, using Student's t test). B, Brg1 ATPase activity was assayed in the presence of GST-⌬C-LKB1.

LKB1 Regulates Brg1 32416
pool. The pools were transcribed and translated in vitro in the presence of [ 35 S]methionine and incubated with either GST or a GST fusion protein of LKB1, bound to GSH beads. The beads were washed, and the bound proteins were separated by SDS-PAGE and visualized with a phosphoimager. We identified a pool that contained a 35-kDa protein that bound specifically to LKB1. By progressively subdividing this pool a single clone was isolated (Fig. 1A). The clone was sequenced, and the 945-base pair cDNA was found to code for amino acids 1080 -1395 of Brg1, containing the helicase-and retinoblastoma (Rb)-binding domains (Brg1HR).
To determine whether LKB1 and Brg1 interact in vivo, LKB1 was immunoprecipitated from Saos-2 cells, and the immunoprecipitates were Western blotted for Brg1. Brg1 was present in the LKB1 immunoprecipitates, indicating that the endogenous proteins associate (Fig. 1B). To identify the region necessary for binding LKB1 we made truncations of Brg1HR (containing amino acids 1080 -1395 of Brg1) that expressed just the helicase domain (Brg1H; amino acids 1103-1231) or the region lacking the helicase domain (Brg1⌬H; amino acids 1231-1395). The Brg1 truncations were transcribed and translated in vitro with [ 35 S]methionine, followed by incubation with GST-LKB1 or GST alone. LKB1 bound to the helicase domain of Brg1 but did not bind to the protein lacking the helicase domain ( Fig. 2A). N-terminal (⌬N-LKB1, containing amino acids 240 -430) and C-terminal (⌬C-LKB1, containing amino acids 1-146 of LKB1) truncations of LKB1 were used to identify the region of LKB1 necessary to bind to Brg1. The constructs were expressed in bacteria as GST fusion proteins, and binding to [ 35 S]methionine-labeled Brg1HR was determined. ⌬C-LKB1 bound to Brg1, but ⌬N-LKB1 did not, indicating that the N terminus of LKB1 is both necessary and sufficient to mediate binding to the helicase domain of Brg1 (Fig. 2B).
To determine whether LKB1 mutants found in PJS bind to Brg1, we transfected Saos-2 cells with HA-tagged wild type LKB1 or the SL26 mutant, and HA immunoprecipitates were Western blotted for Brg1. The SL26 mutant associated with Brg1, as well as wild type LKB1, indicating that the lack of association of LKB1 with Brg1 is not responsible for PJS (Fig.  3A). We also determined whether the kinase activity of LKB1 is necessary for association with Brg1. The SL26 mutant has been reported to lack kinase activity; however this finding is controversial (15,16,18), so we assayed the activity of the SL26 mutant. We did not detect autophosphorylation in either immunoprecipitates from transiently transfected cells (data not shown) or from recombinant GST-SL26 produced in insect cells (Fig. 3B). Furthermore, K252a, a kinase inhibitor that blocks autophosphorylation of LKB1, had no effect on the association of [ 35 S]Brg1HR made in reticulocyte lysate with LKB1 produced in insect cells (data not shown). Thus, the kinase activity of LKB1 is not necessary for association with Brg1.
Because the activity of Brg1 is regulated by phosphorylation (23, 40) we used Brg1 complexes purified from mammalian cells or recombinant Brg1 expressed in insect cells as substrates for LKB1. We detected no increase in phosphorylation above background, with the exception of LKB1 autophosphorylation, in samples to which LKB1 was added (data not shown). These results indicate that the kinase activity of LKB1 is not likely to regulate Brg1-containing SWI⅐SNF complexes directly.
Because LKB1 did not phosphorylate Brg1, we investigated whether its association with LKB1 might regulate Brg1 by measuring the effect of LKB1 and SL26 on the ATPase activity of recombinant Brg1. DNA alone stimulated Brg1 ATPase activity 3-fold, as expected (Fig. 4A). In the absence of DNA, LKB1 or SL26 also stimulated Brg1 ATPase activity 3-fold. In the presence of both DNA and LKB1 or SL26, Brg1 ATPase activity was stimulated 6-fold, compared with Brg1 alone (Fig.  4A). LKB1, SL26, or GST alone did not hydrolyze ATP in the absence of Brg1 (data not shown). These results indicate that LKB1 binding stimulates Brg1 function. We also tested whether the GST fusion protein containing only the N terminus of LKB1 activated the ATPase activity of Brg1 (Fig. 4B). Even though this protein binds to Brg1, it did not enhance Brg1-ATPase activity, indicating that the C terminus of LKB1 is required to stimulate the ATPase activity of Brg1.
Both LKB1 and the SL26 mutant stimulated the ATPase activity of Brg1, raising the question of whether loss of this function could account for PJS. It is not known whether mutant LKB1 proteins are expressed in patients with PJS; if the mutant proteins are not stable, lack of binding to Brg1 could explain the manifestations of PJS. However, the SL26 mutant is stable when ectopically expressed and therefore may be present in the cells of patients with PJS. Furthermore, several mutations in LKB1 found in patients with PJS are point mu-

LKB1 Regulates Brg1 32417
tants that would be expected to eliminate the kinase activity but not disrupt the structure (16). Thus, it is likely that the lack of LKB1 kinase activity contributes to or causes PJS. The ability of LKB1 to induce a G 1 cell cycle arrest in G361 melanoma cells (18) and the association of LKB1 with Brg1, which is necessary for Rb-induced cell cycle arrest in both G 1 and S phases (38,39,41), suggested that LKB1 might function in the Brg1/Rb signaling pathway to induce growth arrest. To test this idea we used the kinase-dead SL26 mutant as a dominant negative to determine whether it would block Brg1dependent growth arrest in SW13 cells, which do not express Brg1. Expression of Brg1 in SW13 cells leads to the formation of large flat cells that have undergone Rb-dependent growth arrest (33,35,36,38,39,42). We determined by Western blotting that SW13 cells express LKB1 (data not shown), thus the SL26 mutant could act as a dominant negative to block endogenous LKB1 function in these cells. SW13 cells were transiently transfected with Brg1, LKB1, SL26, or vector alone or co-transfected with Brg1 and LKB1 or SL26. The cells were selected for 14 days in neomycin, and the number of flat cells per 40 randomly selected 40ϫ fields were counted. Brg1 induced the formation of flat cells, as expected. Co-expression of LKB1 and Brg1 did not affect the number of Brg1-induced flat cells (Table I). However, co-expression of SL26 and Brg1 reduced the number of flat cells ϳ3-fold compared with Brg1 alone (p Ͻ 0.001). Because SL26 lacks protein kinase activity, but binds to Brg1 and stimulates its ATPase activity, these results indicate that the protein kinase activity of LKB1 is likely required for Brg1-induced growth arrest. Co-expression of Brg1 and SL26 did not affect Brg1 expression in transient transfection assays (data not shown). The expression of either LKB1 or SL26 alone induced a 2-fold increase in flat cell formation compared with expression of vector alone (Table I).
LKB1 appears to have both kinase-dependent and -independent functions involving Brg1. Both wild type LKB1 and the SL26 mutant stimulate the ATPase activity of Brg1, but the kinase activity of LKB1 is required for Brg1-dependent growth arrest. Both of these activities of LKB1 may be important in normal cellular function. Expression of RNA coding for a splice variant of LKB1 that lacks a portion of the kinase domain has been identified in several tissues (43). This splice variant, if expressed as a protein, could serve as an endogenous inhibitor of LKB1 function, whose expression, like SL26, would allow cell cycle progression. Our results place LKB1 in a pathway in which Brg1 functions to regulate cell cycle progression. Brg1 heterozygous mice develop epithelial tumors (44), indicating that Brg1, like Rb and LKB1, is a tumor suppressor, and mutations in hSNF5/Ini1, a component of SWI⅐SNF complexes, cause malignant rhabdoid tumors (45). It is likely that defects in cell cycle regulation involving SWI⅐SNF complexes explain the manifestations of PJS, including the predisposition to cancer, caused by LKB1 mutations.